21st April 2020

The world is facing a huge challenge – how to slow the spread of COVID-19. To do so, we need to deploy resources and target interventions in the right places at the right times. A key part of our armoury is a test, or tests, for the causative virus, SARS-CoV-2 and immunity that is acquired following exposure. Several countries have shown how extensive testing for infection can be used effectively to isolate cases early and slow the spread. We can also use these tests to understand when and how to relax (or tighten) lockdown measures by understanding better how the disease spreads (its epidemiology) and by understanding who has been infected and recovered.

In this article, I’ll explain how tests for SARS-CoV-2 work, and why widespread testing is such a challenge.

Firstly, let’s consider what a ‘good’ test looks like. For a ‘good’ test, samples should be straightforward to acquire, and the test should be quick, with a high predictive value. No test is perfect, but we need a test that has a high probability of being right. Ideally, we want few false positives (people who test positive but don’t have the virus) so we don’t needlessly isolate people, such as carers, doing essential work. More importantly, though, we want few false negatives (people who test negative but actually have the virus) as we want to avoid infected care workers treating vulnerable people. A ‘good’ test is one with high sensitivity: that is a test that returns few false negatives. As the Chief Medical Adviser to the UK government, Prof. Chris Whitty, said, ‘the one thing that is worse than no test is a bad test’.

We might also consider a ‘good’ test to be one that can be performed outside of a hospital or laboratory facility, and by almost anyone. Or alternatively by the bedside of a patient for a rapid result (known as point-of-care tests).

So, what do these tests actually look for? Usually, viral tests rely on detecting the presence of one of three molecules: viral genetic material (nucleic acid), viral proteins, or human proteins called antibodies. As the first two are looking for parts of the virus, they allow us to detect if an individual is currently infected. The third can tell us if someone has previously been infected – antibodies are a sign that the immune system has responded to a specific disease.

Viral nucleic acid is normally detected using a lab-based test called quantitative polymerase chain reaction (qPCR). This reaction works by enzymatically replicating specific sequences to detectable levels.

Viral proteins and human antibodies can be detected using tests called immunoassays that take advantage of the fact that certain biological molecules bind very tightly and specifically together (the name reflects the fact they use aspects of an immune response). Viral proteins can be detected with cognate antibodies, and human antibodies can be detected with inactivated chunks of virus. Some immunoassays can be performed by anyone in their own home as they work in a similar fashion to pregnancy tests.

What makes widespread testing so difficult?

Here I will describe four key reasons:
1. It is not always obvious who to test.
2. Nucleic acid tests are fast to develop, but usually resource-intensive to run.
3. Point-of-care tests are quick to run, but resource-intensive to develop.
4. There is a trade-off between tests that are convenient and tests that are accurate.

I will address each reason in turn:

1. It is not always obvious who to test. Tests are most valuable for preventing spread if we can detect infection before someone is contagious. A contagious individual spreads the virus by a process called ‘shedding’. Evidence suggests people with COVID-19 can shed the virus before they show symptoms. This makes it very hard to know who to test as almost anyone could be infectious. By contrast, SARS, another coronavirus-caused disease, was much more controllable because SARS patients do not produce peak viral shedding until a few days after they display symptoms. Anyone presenting a fever could be isolated until they had been tested for the virus – this hugely reduced transmission and meant that SARS never developed into a pandemic.

2. Nucleic acid tests are fast to develop, but usually resource-intensive to run. qPCR-based quantification of nucleic acid is now a well-established technique. The genetic code for SARS-CoV-2 was identified very quickly, so it was relatively easy for researchers to design the reagents needed to run a specific test for the virus. Of course, choosing the best reagents and validating the test takes time, but this is still relatively quick to do. qPCR tests can be highly sensitive, but they usually take 4–6 hours to run and require lots of reagents, lots of trained scientists, and lots of precision machines. There have been challenges in scaling-up manufacture and in distributing reagents. These tests are also usually performed on material collected using a nasal swab, an unpleasant process that requires a trained healthcare worker to perform properly. If this sample is not collected properly, there is a large risk of a false negative in the test.

3. Point-of-care tests are quick to run, but resource-intensive to develop. The best candidates for point-of-care tests are immunoassays – as with pregnancy tests, results can be obtained in minutes, by anyone, anywhere. However, they are more resource-intensive to develop. Immunoassays for viral proteins (to test for infection) require the identification of the best antibodies that bind specifically to SARS-CoV-2. We have the technology to do this, but like making a vaccine, it takes time – it is not unlike finding a needle in a haystack. Immunoassays for human antibodies (to test for prior infection) are a little easier to develop, but still require research efforts to isolate and immobilise viral proteins in a way that they can still be recognised by antibodies in a specific way. The FDA in the US has recently granted authorisation for one of these tests.
There are other options for rapid tests, some of which have been authorised, or are nearing authorisation. In each case development is slow due to the need to ensure specificity.

4. There is a trade-off between tests that are convenient and tests that are accurate. Point-of-care tests usually have a lower accuracy than those performed in a lab. At low current infection rates at a population level, even the most accurate point-of-care tests may have very poor predictive value. More accurate, lab-based tests can be used to verify the results of the convenient, point-of-care tests.

As my colleague G described in a recent blog, “innovation in extreme times”, the pandemic will likely be a catalyst for innovation in diagnostics. New technologies like CRISPR-Cas might enable tests that combine the convenience of a point-of-care test with the accuracy of a lab-based one. Regulatory bodies are adapting to balance acceptable safety requirements with the speed required to make a difference at scale. Funding organisations are thinking about better delivery of money to scientists to help faster development of scientific solutions to societal problems caused by COVID-19. Faster, more accurate, and more accessible tests will help us to slow the spread of this virus and enable us to respond more quickly future pandemic threats.

References
Sheridan, C. “Fast, portable tests come online to curb coronavirus pandemic.” Nature biotechnology (2020).
Wang, C. Jason, Chun Y. Ng, and Robert H. Brook. “Response to COVID-19 in Taiwan: big data analytics, new technology, and proactive testing.” JAMA (2020).
Wilder-Smith, Annelies, Calvin J. Chiew, and Vernon J. Lee. “Can we contain the COVID-19 outbreak with the same measures as for SARS?.” The Lancet Infectious Diseases (2020).
Broughton, James P., et al. “Rapid Detection of 2019 Novel Coronavirus SARS-CoV-2 Using a CRISPR-based DETECTR Lateral Flow Assay.” medRxiv (2020).
https://abbott.mediaroom.com/2020-03-27-Abbott-Launches-Molecular-Point-of-Care-Test-to-Detect-Novel-Coronavirus-in-as-Little-as-Five-Minutes
https://www.medscape.com/viewarticle/928150
Humanity tested. Nat Biomed Eng (2020).